Carlos Cardeira

A low cost mobile robot for engineering education

Robotic competitions are becoming widely used on education. The use of products from the market makes building robots possible at a very low-cost awakening the engineering nature of students and researchers. In this paper we present a low cost mobile robot which uses a normal laptop to guide itself through a track. The low cost robot involves the use of one or more cameras to control the position of the robots position relative to the environment. Motors are from battery operated screwdrivers and the interface with the laptop is based on an USB card, recently made available from a data acquisition cards manufacturer. The goal for this work is to build a large set of these robots, lend them to the engineering students and make them participate in a local university contest. Moreover, along the years, the student integrates more and more knowledge in the robot allowing it to perform as better as their knowledge increases along the course.

Trends in Intelligent Manufacturing Systems

A lot of changes in the manufacturing sector will occur in the future and it is important to transform the actual production systems to evolvable production systems. Fractal, bionic and holonic manufacturing systems are three concepts that have been proposed due their characteristics of flexibility and intelligence. This generation of manufacturing systems is known as intelligent manufacturing systems (IMS). Agent-based software is a technology that can make actions of control or supervision, endowing mechatronics devices with some intelligence. The use of multi-agent based software in operation and control of distributed systems is offering new distributed intelligent control functions (cooperation, planning, scheduling) over wired or wireless networked systems. In this paper, underlying principles of the construction of an IMS are concisely presented.

A schedulability analysis of tasks and network traffic in distributed real-time systems

In real-time systems the deadlines of each task must be met. A pre-run-time schedulability analysis becomes necessary to prove that the existing software and target hardware will meet the real-time application constraints. In a real-time distributed system, the messages transmitted through the network are also time constrained. However, some new problems arise when one applies existing task scheduling algorithms to schedule the network traffic. The main goal of this paper is to define the boundaries between these two domains of scheduling. After an introduction to fieldbuses and real-time systems, we present an equivalence between tasks and messages as well as between processors and networks, which are much different in practice but have strong similarities from the scheduling point of view. Finally, we analyse the new constraints introduced by the presence of smart transducers/transmitters in fieldbus applications and we define the scheduling algorithms adapted for this type of applications.

Neural networks for multiprocessor real-time scheduling

In recent years, neural networks have become a popular area of research, especially after Hopfield and Tank opened the way for using neural networks for optimization purposes and surprised the scientific community by their paper (Biological Cybernetics, vol. 52, pp. 141-52, 1985) presenting a circuit to give approximate solutions for the classical traveling salesman problem in a few elapsed propagation times of analog amplifiers. In this paper, we analyze Hopfield neural networks from the scheduling viewpoint to see if they can be used to solve real-time scheduling problems. We build a neural network whose topology depends on real-time task constraints, and converges to an approximate solution of the scheduling problem. Finally, we analyze the quality of the result in terms of the convergence rate and the complexity of the algorithm.<<ETX>>

An open architecture for position and force control of robotic manipulators

In this paper a low cost open architecture PC-based axis controller for robotic applications is described. The system is based on an available low price commercial PC interface board and the controller software runs totally on a Pentium 100 MHz computer in order to control a robotic axis. The driving kernel of software for supervisory, management and control was developed in a high level language (C++) and enables the analysis of several position and force/position control strategies for research and practical applications. The experimental results for the given position and force trajectories reveal promising position and force tracking results for the experienced control algorithms.

Handling precedence constraints with neural network based real-time scheduling algorithms

In previous work, the authors proposed an approach to the approximate solution of scheduling problems, neural network based algorithms, applied to the preemptive and non-preemptive scheduling for a mono or multiprocessor environment. Results were presented in a systematic approach for translating task constraints into neural network building rules that are independently added to the neural architecture. The main advantage of this methodology is that the neural network built according the rules converges to a solution of the scheduling problem in only a few propagation times of analogue amplifiers. They present new rules that extend the methodology to handle precedence constraints. They present the formal energy function which occurs when the precedence constraints are met and finally present a performance analysis of the quality of the results obtained by this approach.

Solving real-time scheduling problems with Hopfield-type neural networks

Real-time applications are increasingly becoming more complex, leading to the necessary development of fast scheduling algorithms. Therefore, the use of algorithms with a parallel search of feasible schedules seems to be attractive. In turn, Hopfield-type neural networks are suitable to solve complex combinatorial problems, owing to their fast convergence, if analog hardware is implemented. However, these neural networks have associated concepts of sub-optimality and the possibility of unfeasible solutions, which are contrary to the notion of system predictability. The paper presents a systematic procedure to map the scheduling problem onto a neural network in such a way that network solutions are always feasible schedules. Network convergence time is studied with digital computer simulations, using a discrete time model. Global asymptotic consistency between the discrete time model and the continuous one is assured. The paper also presents an analysis of the complexity of the proposed method.

2D PCA-based localization for mobile robots in unstructured environments

In this paper a new PCA-based positioning sensor and localization system for mobile robots to operate in unstructured environments (e.g. industry, services, domestic...) is proposed and experimentally validated. The inexpensive positioning system resorts to principal component analysis (PCA) of images acquired by a video camera installed onboard, looking upwards to the ceiling. This solution has the advantage of avoiding the need of selecting and extracting features. The principal components of the acquired images are compared with previously registered images, stored in a reduced onboard image database, and the position measured is fused with odometry data. The optimal estimates of position and slippage are provided by Kalman filters, with global stable error dynamics. The experimental validation reported in this work focuses on the results of a set of experiments carried out in a real environment, where the robot travels along a lawn-mower trajectory. A small position error estimate with bounded co-variance was always observed, for arbitrarily long experiments, and slippage was estimated accurately in real time.

Towards Ubiquitous Production Systems and Enterprises

In the evolvable manufacturing world we are moving towards systems made of small intelligent interacting units which make their own decisions independently from a centralized decision maker. System intelligence will be the collective intelligence resulting from a large number of individual decision, each one based on its local knowledge. in our paper we present the ubiquitous production systems and enterprises (UPSE) whose foundations can be found in two main scientific areas/disciplines, (1) ubiquitous computing systems and (2) virtual and networked enterprises and organizations. We present the concept and the underlying formal model as well as our positioning in the state of the art of similar concepts. Finally we present some ongoing experiments that will work as a testbed for the UPSE concept.

Compu-search methodologies II: Scheduling using genetic algorithms and artificial neural networks

This chapter complements the previous chapter ‘Scheduling Methodology: Optimization and Compu-search Approaches I’ about the scheduling level of a production manufacturing hierarchical approach. It presents various ways of using genetic algorithms and artificial neural networks to solve scheduling problems. Genetic algorithms are used for scheduling problems without assignment unknown values (solutions are completely decribed by the list of job sequences on each resource). The potential use of artificial neural networks for solving scheduling problems is illustrated with a simple multiprocessor scheduling problem.